The Utilizing Predictive Analytics to Reduce Time-to-Market for Consumer Packaged Goods

Abstract

The consumer packaged goods (CPG) business is known for its fast changing consumer preferences, stiff competition, short product life cycles, volatile promotions, supply chain uncertainty, and high expectations for the development of innovative products on the fast track. The increased importance of time-to-market has led to delays in product development, demand forecasting, procurement, manufacturing, packaging, regulatory clearance and commercialization becoming critical determinants of competitive advantage in product development, resulting in significant development costs if companies miss market opportunities. This study sought to explore how predictive analytics can be used to shorten the time to market in CPG, combining demand forecasting with machine learning, predictive project-risk metrics, supplier performance metrics, product complexity and market intelligence into one analytical framework. The data was gathered from 35 consumer packaged goods product development initiatives and related operational records for product-development cycle time, demand forecasts, historical sales, supplier lead times, product complexity, marketing activity, production readiness, regulatory requirements, and outcome of the product commercialization. Descriptive statistics, correlation analysis, multiple regression analysis, variance inflation diagnostics, and machine-learning analysis were used in the empirical analysis. A comparison between the Random Forest, Gradient Boosting and Extreme Gradient Boosting (XGBoost) models and standard multiple regression and time-series forecasting methods was performed. The performances of the models were assessed by the mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE) and coefficient of determination (R²). The results showed that there is a significant negative relationship between product-development cycle time and predictive analytics. There was a correlation between development and commercialization time and higher predictive-analytics capability. Some of the key drivers of time-to-market were identified as demand forecast accuracy, supplier lead-time predictability, production readiness prediction and early identification of project risks. The machine learning models performed better than the conventional regression in predicting project cycle time, and the XGBoost model gave the best overall predictive performance. The analysis also showed that predictive analytics could reduce uncertainty in decisions relating to demand and supply and could help identify activities earlier that could cause delay in commercializing the innovation. The results confirm the idea that predictive analytics shouldn't be looked at as a forecasting tool only, but as a cross-functional decision support function, encompassing market information, product development, ordering, production, and commercialisation. The study adds to the literature by associating predictive analytics with time-to-market, rather than focusing on forecasting demand and supply-chain performance. The results have implications for CPG managers who want to shorten the time to market, get the cross-functional team working together more efficiently, cut down on unnecessary delays and boost the likelihood of a product's success.